Abstract The research proposed here is in the area of cancer predisposition genetics, and its focus is on the moderate- risk cancer susceptibility genes (ATM, CHEK2, etc.) in the biochemical pathway responsible for DNA double strand break homologous recombination repair (HRR). Most of the genes in the HRR pathway are breast cancer susceptibility genes, with several also being ovarian cancer susceptibility genes and/or pancreatic cancer susceptibility genes. In the first funding cycle, we discovered that the majority of pathogenic alleles in the moderate-risk HRR genes are individually rare missense substitutions (rather than the individually rare protein truncating variants that dominate the mutation spectrum of BRCA1 and BRCA2). From a clinical cancer genetics and patient counseling point of view, this observation creates a serious problem: the clinical testing labs typically report these rare missense substitutions as Variants of Unclear Significance (VUS). Because the VUS are not used for patient counseling, this means that the majority of the bona fide genetic risk detectable in the moderate-risk HRR genes by the panel tests is not used for patient counseling and risk management! Over the last 12 years, we played a central role in development of methods for clinical classification of VUS in BRCA1 and BRCA2, and are currently developing corresponding methods for the colorectal (and other cancer) susceptibility genes MLH1, MSH2, PMS2, and MSH6. Here, we hypothesize that combining improved computational methods for rare missense substitution evaluation with comprehensive high-throughput functional assays will dramatically accelerate the process of evaluation and classification of clinically observed VUS. Thus the first Aim of the project focuses on confirming and then improving the accuracy of computational methods for evaluation of rare missense substitution. The second Aim is directed towards development of medium-to-high throughput assays of missense substitution functionality. In terms of genes, we will begin these studies with the RING domain of BRCA1 (which lacks a properly calibrated functional assay), and then progress to ATM and then CHEK2. In the third Aim, we re-calibrate the outputs from the computational methods of Aim 1 and the functional assays of Aim 2 into probabilities (or odds) in favor of pathogenicity, the key variables required for clinical-quality classification of sequence variants observed in people. We then combine these data to obtain posterior probabilities in favor of pathogenicity and pass those posterior probabilities through a well-recognized categorical classifier (the IARC standards) to generate classification recommendations for clinicians and patients. Progress across the three Aims of this study will dramatically accelerate variant classification while simultaneously improving the sensitivity and precision of the clinical testing process.